Haojun Zhao
Haojun Zhao
Personally, I had this problem because I used python 3. Maybe make sure that you use python 2 and then the directory to ontonotes dataset is correct.
Please wait a little bit to execute tests, I still need to change the nanotron code. Already made a PR to nanotron
Thank you for confirming the issue @calvintwr
Same question. Did you solve it? Also in switch aux loss, why we don't need to do all gather for 'probs'?